Arrive with an idea, leave with a 3-year plan. Achieve reliability.

TRC gives you access to cutting-edge knowledge & technology

Sign Up

Please use your business email address if applicable

2013 Uptime Award Winner: If Reducing Unplanned Downtime Is Your Goal, What Is Your Game Plan?


A Case Study

Zellstoff Celgar is a market pulp mill located in southeastern British Columbia and sister mill to two other similar sites in Germany, owned by Canadian-based Mercer International. The mill, by North American standards, is a modern producer of pulp, having been upgraded in 1993 following a C$850 million rebuild project.

The mill produces 520,000 air-dried metric tons (ADMT) per year of northern bleached softwood kraft (NBSK) pulp, a versatile raw material used to make many finished grades of paper, including printing, writing and tissue. The mill consists of three fundamental parts: the fiber line, including wood fiber, cleaning and screening, cooking and washing, bleaching and chemical recovery, including causticizing; the pulp drying, baling and shipping; and the utilities area, including combined heat and power (CHP) and water and effluent treatment.

Converting wood fiber to market pulp is a chemical process involving several interconnecting processes, each with hundreds of assets, many of which rotate. A significant part of the process is CHP generation. Excess power, in the form of electricity, is sold to the local utility at competitive rates, generating a separate, but very important revenue source for the mill.

Of 8,657 rotating pieces of equipment in the mill, the predictive maintenance (PdM) team monitors 856 assets based upon criticality. Criticality is based on the asset’s contribution to the site’s business goals and changes with business conditions. An ongoing, periodic maintenance strategy review (MSR) using reliability centered maintenance (RCM) provides updated priorities and strategies.

Condition Monitoring Program

The current condition monitoring program consists of over 50 wireless units, several fixed online data collection stations and eight mobile data collection units configured for use by operators and vibration technicians.

The turbine generator #2 (TG2) and turbine generator #3 (TG3) have protection systems. Both are also equipped with a condition monitoring system. The TG2 protection system is dated and is planned to be replaced in 2014. There are also plans during 2014 to further expand the mill’s condition monitoring program with four online data collection stations.

Program Expectations

The reliability effort at Zellstoff Celgar is tasked with these expectations of the PdM program:

  • Targeted 92 percent uptime. This is a moving target depending on product mix, market conditions and business goals. For example, in 2006, production was 1283 T/day, while in 2011 that increased to 1520 T/day.
  • Maintenance cost versus replacement asset value (RAV) goal of three percent.
  • Safety total incident rate (TIR) of 1.76.
  • Zero failures on criticality “1” equipment.

Changing Priorities

The second steam turbine (54 MW) installed in 2010 changed the mill’s production strategy, balancing priorities between production of pulp and sale of electricity. This change required an adjustment to the existing asset criticality and corresponding maintenance strategies.


The original program consists of two certified vibration analysts. In addition to the vibration technicians, there are seven certified lubrication technicians, supervised by the PdM group supervisor, who has a bachelor’s degree in mechanical engineering and is a certified vibration analyst. Included in the reliability effort are a motor condition analysis technician and two reliability engineers.

The program’s effectiveness is directly related to the efforts of this group of skilled, knowledgeable, dedicated and focused specialists to constantly try to improve the mill’s reliability.

Program Effectiveness Issues

An evaluation of the traditional PdM program’s effectiveness reveals several issues:

  • Dynamic production processes with continuous changes and updated automation requires new knowledge and skills.
  • Skilled people’s retention and increasing average age of operations and maintenance professionals, plus the expectation to do more with less.
  • The technicians can collect data, but there is no process for using the data effectively and efficiently.
  • Inconsistent data analysis efforts caused by different personnel analyzing data with different levels of knowledge and experience.
  • No continuity of the maintenance effort, therefore, history is lost with no effective means to improve.

This classic approach to a PdM program has problems:

  • Route-based, taking readings every two or four weeks instead of as required depending on condition.
  • Collecting data before and upon start-up after shutdowns and for all new asset installations strains the routine data collection.
  • Insufficient troubleshooting expertise to determine how bad is bad and how long before intervention is required.
  • Data analysis is time demanding, largely subjective and relies upon single technology (e.g., no interaction with process data or different operating parameters).
  • The analysis is highly dependent on the analyst’s skills, the quality of the database, the point setups, alarm adjustments, the process for tracking changes and the instrument type.
  • Even after a problem is found, providing a solution based on someone’s knowledge and experience, entering work requests, executing the corrective action and following up has no effective means to improve.


An @ptitude Decision Support system (@DS from SKF) was installed as part of the technology upgrades. The system does all the diagnostics, providing recommendations, supported with references, through e-mail notifications.

A certified vibration analyst evaluates all system recommendations and makes the call whether to adjust alarms for that specific asset, do more troubleshooting, or initiate a work request.

The support system is the main asset health diagnostic center, continually providing diagnostics for 1,472 assets at 30 minute intervals, providing not only alarm notifications, but also recommended actions. This system processes huge amounts of data, some 200,000 different readings a day, while not requiring database growth. The system closely monitors critical equipment by analyzing vibration data, process parameters, oil analysis results and operator driven reliability (ODR) program inspection data. Critical equipment is monitored before shutdown and during start-up, reducing resource constraints.

The mill’s root cause analysis (RCA) program is supported by the system providing faults statistics. This same support is critical for those so called “big calls” that impact production, safety and the environment.

Configuration of the system uses existing models provided by the vendor, plus development of the mill’s models based on the function of the assets in their operating environment. Additionally, adjustment of existing models to reflect existing circumstances is a continuing process.

This development and configuration took place over a six year period as assets were added to the program and technology was purchased and installed to collect data. It is estimated that 400 to 500 assets can be modeled into the application in three to four months and will require up to one year of adjusting or tweaking, providing near 100 percent accuracy.


First, the database is optimized (e.g., proper point setup, alarming and faults identification). Optimally, this optimization should take place periodically, as assets and priorities change regularly.

Next, designed standard models for motors, pumps, gearboxes, process rolls, mixers, fans, etc., are incorporated into the application. Examples include electric motors, roller bearings, sleeve bearings, AC variable frequency drives and DC variable frequency drives in 900, 1200, 1800 and 3600 RPM.

Each potential fault must be connected to the specific data providers, for example, route-based data, wireless, ODR and process data from the distributed control system (DCS) through OPC. Then, each asset’s profile is tested, making sure alarms are properly set and data providers connected correctly.


E-mail notifications, via desktop, smartphone and/or tablet, is the primary communication. The notification includes a recommended corrective action and the degree of severity. An interface between @DS and the new computerized maintenance management system (CMMS) is being developed to create work requests and work completion notifications.

Implementation Challenges

Some of the challenges during implementation were due to the fact that most decision support models were generic and required customization to site and process-specific circumstances. In addition, 90 percent of the models had to be rebuilt and fine-tuned in-house, a very slow and time-consuming process.

A duplicate e-mail notification problem has recently been resolved. Currently, managing alarms requires at least a Level 2 analyst, but this is a good training exercise leading to analysts becoming experts and learning about rules and models.

Currently no interface exists between this system and the CMMS, although development is in progress. The work request has to be entered manually, requiring time from a resource. Checking the CMMS for duplicate requests or work orders is also required.

Valves are a critical asset in pulp and paper mills and currently there are no developed models for monitoring valves, so the mill is in the process of reviewing an interface.

Zellstoff Celgar’s Expectations

The goal is zero unpredicted failures on all 854 critical monitored assets under the PdM program. Additionally, continuous improvement is to be measured by mean time between failures (MTBF), especially on pumps and electric motors.

The establishment of key performance indicators (KPIs) benchmarks resulting from a site assessment led to a development plan to close the identified gaps. This ensures that the continuous improvement process is in place.


Several issues have been resolved as a direct result of the program.

  • TG2, 48 MW Steam Turbine Shutdown Call:
    • The mill was able to add this overhaul to the planned 2010 shutdown due to predicted failure issues, thus mitigating a potential catastrophic event. Originally, the planned overhaul was scheduled for 2011.
  • Pump Cavitation:
    • Effluent pumps performance can limit production; one pump down a 20 percent production rate reduction in order to meet regulations is experienced. This rate reduction averages 1,500 tons of pulp until corrected.
    • A bottom circulating pump on the digester was damaged during a planned shutdown, however, the damage was discovered during startup. Repairs delayed start-up by eight hours, but a catastrophic failure that could have resulted in potential loss of production for 24 hours was avoided.
  • Shutdown and Start-up Problems:
    • A bleach plant do stage feed pump electric motor fault was noted on startup. A simple addition of grease avoided a possible 400HP motor failure and start-up delay. Without the support system and the wireless network as a data provider, there would have been another failure and RCA process candidate.
  • Processing Huge Amounts of Data:
    • The resource level, for instance three condition monitoring technicians, could not possibly analyze the huge amount of data being processed continuously. To date, the recommended corrective actions produced by the system have been 100 percent correct.
  • A traditional PdM program without continuous monitoring and automated analysis would have missed the failure shown in Figure 1, a gearbox jamming on the digester outlet device, which allowed a controlled shutdown and correction versus an unplanned, uncontrolled outage.
  • Zero unpredicted downtime on 854 critical assets.



The decision support system enabled the mill to do better by providing instantaneous, reliable diagnostics with useful recommended actions, thus avoiding catastrophic production failure events, lengthy production start-up delays and costly repairs. As a result, the mill is able to devote more resources to RCA and the elimination of repetitive failures.

Mercer International and Zellstoff Celgar’s management are pleased with the results from the decision support system and are in the process of reviewing additional implementation at its two German mills.

The next steps include incorporating valve monitoring analysis in the @ DS; interfacing with the CMMS or enterprise asset management system; and expanding the ODR program from the current four designated areas. Also coming up are the installation of new continuous monitoring systems and a change in the scope of work for vibration analysts to more follow-up analysis from ODR and continuous monitoring systems instead of route-based data collection.


SKF Client Needs Analysis - Asset Management (CNA AM) survey of 1,100 industrial sites.

Uptime Awards -

Keep reading...Show less
ChatGPT with
Find Your Answers Fast